Setup & Installation
What This Skill Does
A guide for building persistent memory into AI coding agents. It covers hook-based capture, two-tier storage (global and project-local), reflection patterns, and compaction rules so agents carry knowledge forward across sessions.
Without this, every agent session starts from zero — corrections, conventions, and failure patterns all disappear, so the same mistakes repeat indefinitely.
When to use it
- Storing project-specific parameter names after a user correction so the agent stops repeating the same mistake
- Surfacing a cached tool failure pattern at session start before the agent tries the same broken approach
- Scoping Azure auth conventions to a repo so global memory stays clean
- Pruning stale learnings automatically while keeping high-frequency entries indefinitely
- Writing SQL inserts directly from agent code to record team preferences without manual intervention